zomony

How Zomony Finds Every Product in a Photo

Most shopping starts with seeing something you like before you know what it's called or where to buy it -- a lamp in a friend's living room, an outfit in a video, a desk setup you screenshotted months ago. Zomony exists to close that gap: upload the photo, and it tells you what's in it and where to look for it.

Here's what actually happens when you upload a photo.

The photo goes to a real AI vision model

Zomony sends your image to Amazon Bedrock, running Anthropic's Claude, through Bedrock's Converse API. This isn't a canned demo response or a pre-scripted answer -- it's a real vision model looking at the actual pixels of your actual photo.

That matters because it's the difference between a system that can only recognize a fixed list of pre-trained categories, and one that can look at your desk setup, in your lighting, in your framing, and reason about what's actually there.

It identifies distinct products, not just "the scene"

A lot of image recognition stops at describing the scene as a whole -- "a desk with some items on it." Zomony is built to go further: it picks out the individual, distinct products in a photo. Upload a photo of a desk setup, and instead of one vague label, you get the keyboard, the monitor, the desk lamp, and the mouse pad identified as separate items, each one something you could actually search for on its own.

This is the core design goal, and it's also the hardest part -- a cluttered photo has a lot going on, and the model has to separate "distinct product" from "background object" from "part of another object."

Each item gets a real place to buy it

Recognition alone isn't useful if it doesn't lead anywhere. For every product Zomony identifies, it builds a real search link to a marketplace -- currently Shopee -- so you land on actual current listings, not a static or fabricated catalog. You're not looking at "a lamp like this exists somewhere"; you're looking at real sellers, real prices, right now.

Where this is headed

Zomony today handles the core loop -- upload, identify, find where to buy -- and does it with a real model and real marketplace links. What's still ahead: more marketplaces beyond Shopee, saved collections so you can come back to what you've found, and richer product matching. We'd rather ship the core loop well and tell you honestly what's next than promise features that don't exist yet.

If you want to see it for yourself, upload a photo and see what it finds.